Late identification of water leaks can cause significant damage to buildings and infrastructure, leading to costly repairs and wasting resources. Traditional methods of detecting water leaks from pipes can be highly time-consuming and may not provide real-time information, which is key to identifying leaks early and taking appropriate action. This paper is about to address these issues by developing an ultrasonic water leak detection system with real-time transmission of measured values. The proposed system uses ultrasonic transducers strategically placed in the water supply network to detect water leaks by measuring changes in ultrasonic signals. The measured values are processed and transmitted in real-time using a microcontroller with a wireless communication module that allows remote monitoring of the system. In order to increase the accuracy and predictive capabilities of the water leak detection system, this paper also incorporates machine learning techniques. The predictive models are integrated into the system, allowing it to proactively anticipate and alert on potential leaks before they develop into larger problems.